Geodetic GPS-Based Evaluation of Present-Day Geodynamical Movements and Geoid in the Nares Strait Region

Polarforschung 74 (1-3), 161 – 168, 2004 (erschienen 2006)

by Mette Weber1, Anna B.O. Jensen2*, René Forsberg3 and Finn B. Madsen3

Abstract: The Nares Strait region between Ellesmere Island and Greenland wurden mit Hilfe hoch entwickelter Algorithmen der Berner GPS-Software has for many years been of significant geological interest, since the details of unter Berücksichtigung präziser Satelliten-Bahninformationen, Erdgezeiten- the plate tectonic evolution of the region is still being debated. Even though modellen etc. prozessiert. Die auf den beiden Kampagnen gemessenen Posi- the region is currently seismically inactive, it is still relevant to investigate tionen wurden analysiert, um wenn möglich deutliche laterale Bewegungen whether the area is presently affected by movements, and if so, how large innerhalb dieser letzten sechs Jahre nachzuweisen. Die vertikale Komponente these movements are. Movements of interest are both the horizontal changes wurde verglichen mit einem modernen geodynamischen Hebungsmodell. (if any) and – especially – the vertical uplift (the Nares Strait region is a major uplift anomaly in current glacio-isostatic rebound models). With the develop- ment in GPS, monitoring of geodynamical changes are a major focus point for geodesy – the traditional geodetic discipline that describes the size and the INTRODUCTION shape of the Earth. Repeated measurements in high-accuracy satellite geodetic reference networks are used for this, and any movements of the reference The Nares Strait separates Ellesmere Island in the Canadian points, caused by geodynamic activity, must be considered in defining the reference systems. Arctic from Greenland. The exact nature of the plate-tectonics This paper describes two geodetic GPS survey campaigns carried out in the of the region is still being debated. Concerning present-day Nares Strait region in 1995 and 2001, repeating observations in selected geodynamic movements, modern geodetic GPS precise posi- points from the classical joint Danish-Canadian triangulation along the Nares Strait completed in the early 1970’s. Both recent GPS surveys were “opportu- tioning techniques have reached an accuracy level down to a nity missions” relying on logistics provided either by gravity survey activities few mm so that such methods provide an excellent tool for the (1995) or the research vessel “Louie S. St Laurent” (2001). Approximately ten investigation of possible present movements of the land, not geodetic markers were surveyed during both campaigns, and the positions of the well mounted markers were determined with accuracies of a few milli- just in the horizontal coordinates (primarily relating to tecto- metre by utilising days of GPS data. In addition the 2001 field campaign also nics) but also in the vertical (related, for instance, to the recent included sea-level observations for geoid control. glacial loading history). The use of permanently mounted GPS The GPS data was processed using advanced processing algorithms imple- stations are preferable for investigations of such geodynamical mented in the Bernese GPS processing software, along with precise satellite orbit information, earth tide models etc. The positions determined from the activity. However, permanent GPS stations do require a consi- two survey campaigns are analysed with the purpose of detecting if any signi- derable amount of logistics and infrastructure, and are there- ficant relative movements have taken place during the six years, and the fore difficult to establish in remote areas such as the Nares vertical movements are compared with a contemporary geodynamic uplift model. Strait region. Zusammenfassung: Die Region der Nares Strait zwischen Ellesmere Island The second best approach is to carry out repeated GPS survey und Grönland hat seit vielen Jahren das geologische Interesse auf sich campaigns, where GPS receivers are carefully mounted at gezogen. Im Detail wird die geologische Entwicklung noch immer diskutiert. Obwohl die Region derzeit seismisch inaktiv ist, lohnt es doch zu prüfen, ob benchmarks and left at the sites collecting data for several und in welcher Größenordnung die Region von rezenten Bewegungen days. With several days of carrier phase GPS data, and by betroffen ist. Von Interesse sind beide Bewegungskomponenten, die horizon- carefully processing the data relative to the global GPS refe- tale (so vorhanden) und vor allem die vertikale, da die Region eine größere Hebungs-Anomalie in aktuellen Modellen postglazial-isostatischer Entlastung rence system, coordinates with accuracies at the sub-cm level umfasst. Der Nachweis von geodynamischen Veränderungen ist seit der can be obtained. If the sites are repeatedly visited, the obtain- Entwicklung der GPS-Methoden ein wichtiges Feld für die Geodäsie, die able level of accuracy is sufficient to detect geodynamical traditionelle geodätische Disziplin, die Größe und Form der Erde beschreibt. Man benutzt Wiederholungsmessungen im Rahmen von hochgenauen activity after a few years. This approach has been followed in geodätischen Netzwerken. Alle durch geodynamische Bewegungen veru- the present case of the Nares Strait, where we analyse results sachten Änderungen der Referenzpunkte müssen bei der Definition der Netze of GPS survey campaigns carried out in 1995 and 2001, both berücksichtigt werden. as “opportunity missions”, with the 1995 campaign measure- In dieser Arbeit werden zwei geodätische GPS-Kampagnen beschrieben, die in der Region der Nares Strait 1995 und 2001 als Wiederholungsmessungen ments being suboptimal, since generally no more than 24 hour der klassischen dänisch-kanadischen Triangulation der frühen 70er Jahre GPS observation sessions were possible due to logistic durchgeführt wurden. Beide Kampagnen waren „Gelegenheits-Projekte“, die constraints. Seven locations were visited during both die Logistik eines Gravimetrie-Projekts (1995) und des Forschungseisbrechers „Louis S. St. Laurent“ (2001) nutzten. Etwa zehn geodätische Referenzpunkte campaigns (Fig. 1), and by processing the GPS data from the wurden in beiden Kampagnen neu bearbeitet. Die Positionen der im Gelände two campaigns in the exact same way, the two sets of position sorgfältig fixierten Messpunkte wurden per GPS mit einer Ge nauigkeit coordinates can be compared, and the movement of the sites weniger Millimeter bestimmt. Die Kampagne von 2001 umfasste daneben can be determined. Bestimmungen des Meeresspiegels zur Kontrolle des Geoids. Die GPS Daten ______In this paper initially the two survey campaigns are described. 1 National Survey and Cadastre (KMS), Rentemestervej 8, DK-2400 Copenhagen, Den- mark. This is followed by a brief review of the GPS processing 2 Informatics and Mathematical Modelling, Technical University of Denmark, Richard theory, and the processing parameter set-up. The results of the Petersen Plads, building 321, DK-2800 Lyngby, Denmark. 3 Danish National space Centre, Juliane Maries Vej 30, DK-2100 Copenhagen Ö, Den- two survey campaigns are compared in order to detect the mark. *Corrresponding author. movements, which have taken place during the six years. The movements are discussed and analysed, and they are compared Manuscript received 07 October 2005, accepted 13 October 2005 with the ICE-4G geodynamical model (PELTIER 1996). We

161 Fig. 1: Stations observed during both the 1995 and the 2001 survey campaigns. finally also use the results of the combination of GPS and tide nent GPS station THU1 at Thule Air Base, and station RES1 in gauge data, collected as auxiliary data in 2001, for geoid Resolute Bay, were operated as reference stations, conti- control. nuously logging data throughout the field campaign.

THE 1995 FIELD ACTIVITY THE 2001 FIELD ACTIVITY In 1995 the National Survey and Cadastre of Denmark (KMS) participated in the joint Danish-Canadian “Op Bouguer” With the Nares Strait GeoCruise 2001, an opportunity arose to project, a joint venture between Mapping and Charting Esta- revisit the sites observed in 1995, and based on the two diffe- blishment, (Canadian Defence), Canadian Hydrographic rent data sets analyse whether any movements (locally or Service, KMS and NIMA. The main purpose of this project regionally) had taken place over the six years. The Nares Strait was to cover Ellesmere Island and the adjacent parts of Green- GeoCruise was established as a research cruise mainly by land with gravity measurements on a 10-15 km grid, including initiative of the German Bundesamt für Geowissenschaften gravity measurements on the sea-ice of the Nares Strait. und Rohstoffe (BGR), and the Geological Survey of Canada, Bedford Institute of Oceanography. Several other research Using the available helicopter logistics, a GPS survey of the groups participated in the cruise as well. older 1970’s triangulation points along the Nares Strait was done, mainly in order to establish a “zero epoch” for future A total of 15 sites were visited with GPS during the 2001 GPS geodynamic studies. The survey was done by KMS, in coope- campaign, and seven of these sites were identical with sites ration with the Geodetic Survey Division, Natural Resources visited during the 1995 survey. As a part of the 2001 field- of Canada (NRCan), who operated sites at Grise Fjord and work, new and more stable station markers were established at Resolute Bay. The field work and the results of the processing the sites in order to improve the accuracy of any future GPS are documented in FORSBERG et al. (1995). The location of the surveys in the area. The new markers are so-called GPS-bolts, 19 sites observed with GPS are shown in Figure 2. The perma- drilled and glued into the rock. The GPS antennas can be mounted directed on the bolts, whereby any inaccuracies from the antenna setup are eliminated. Data from short observation time spans (c. 30 minutes) between the old and the new markers were collected to be able to position the two markers relative to each other. The GPS fieldwork of the 2001 survey was carried out as a cooperation between the Geodetic Survey Division of NRCan, Asiaq - Greenland Survey, and KMS. More information on the fieldwork is given by JENSEN et al. (2001).

Locations of the GPS sites that were visited during the 2001 survey campaign, and the network used for the GPS data processing are shown in Figure 3. The GPS stations THU2 at Thule Air Base, RES0 in Resolute, GRIS in Grise Fiord and EURK in Eureka were operated as reference stations collecting GPS data continuously during the cruise. The permanent Fig. 2: Network configuration in 1995. station in Thule is an IGS station operated by KMS, and the

162 instance HAN & RIZOS (1997) for a review of some techniques.

The sizes of dρ, dion, and dtrop (the spatially correlated errors) are reduced when performing double differencing, and for vectors shorter than approximately 20 km these errors are generally negligible after double differencing.

For longer vectors the spatially correlated errors do not cancel out, and may therefore be treated as part of the positioning process. For post processed positioning the error in satellite position can be mitigated by using precise post processed satellite orbit information for instance from the IGS (NEILAN et al. 1997).

The ionosphere is dispersive for radio waves, so the ionos- Fig. 3: Network configuration in 2001. pheric effect is different for the two GPS frequencies referred to as L1 and L2. When working with dual frequency receivers permanent stations in Eureka and Resolute are operated by which is normally the case for high accuracy positioning, the NRCan. The stations in Thule and Resolute are new installa- ionospheric error can therefore be mitigated by using the so- tions, and thus not the same as the ones used for the 1995 called ionosphere free linear combinations of the L1 and L2 survey. The station in Grise Fjord was a temporary set-up for observations (SEEBER 1993). Hereby the first order ionos- the duration of the 2001 cruise. pheric effects are removed from the positioning process. Higher order effects of the ionosphere will still be present, but for vectors of 50-100 km their influence is relatively small. For GPS DATA PROCESSING THEORY longer vectors, the higher order effects should, however, be considered (BRUNNER & GU 1991). The ionospheric activity is Before the processing procedure and parameters used for the correlated with the solar activity and thus also with the geoma- present GPS data processing are described, a brief summary of gnetic activity. It is therefore possible to use indices for solar the GPS carrier phase based positioning theory is given below. and geomagnetic activity as indicators for the amount of ionospheric activity, and this has also been done in the present When using GPS for high accuracy differential positioning, at case, as described in the following sections of the paper. least two GPS receivers must be used. One receiver is located at a reference point with known coordinates and the other The tropospheric delay is caused by refraction of the satellite receiver (the rover) is located at a point with unknown coordi- signal as it is transmitted through the lowest parts of the Earth nates. The position of the rover is then determined based on atmosphere. The refraction, and thereby the signal delay, is a the vector between the two stations, or in other words relative function of the meteorological conditions along the signal to the reference station. path, and for positioning purposes the majority of the effect is handled by global tropospheric delay models. For high accu- To obtain the high position accuracy both code and phase racy positioning the residual effect after both modelling and observations collected by the two receivers are used. The GPS double differencing can be estimated through the positioning positioning process is based on the phase observation equation process. This tropospheric effect is considerably smaller than as described, for example, in KLEUSBERG & TEUNISSEN (1996). the ionospheric effect and for differential GPS positioning With observations from two receivers and two satellites at the carried out in dry cold conditions, the residual tropospheric same epoch in time, the observation equation can be generated effects are normally not causing any problems in the positio- for each receiver-satellite combination, and the equations can ning process. be twice differenced (double differenced). The result is the double differenced phase observation given in Equation (1): The effect of the remaining error sources i.e. receiver noise and multipath is small because of the high quality equipment ͬͭ ͬͭ ͬͭ ͬͭ ͬͭ ͬͭ ͬͭ Φ = ρ + dρ - dion + dtrop +λ N + ε used, and because the GPS-sites are all located in environ- ments with little multipath. Where ͬͭ is the double difference operator, ρ is the geometric distance between receiver and satellite, dρ is the error in the satellite position, dion is the ionospheric delay, dtrop is the tropos- GPS PROCESSING CHARACTERISTICS pheric delay, λ is the wave length, N is the ambiguity - the initial number of full carrier phase cycles, and ε is the noise In 1995 GPS data were collected from 27 May to 19 June. At term including mainly multipath and receiver noise, but also the time of observation the solar activity cycle was close to a any un-modelled errors. All elements are given in range units. minimum, and the days were all characterized by a low to moderate ionospheric activity, which is not expected to cause The double differenced ambiguity, ͬͭN, is by nature an any disturbance in the positioning process. In 1995 the obser- integer number and to obtain position accuracies of a few cm vation time spans for the stations of interest were 24 to 36 or mm the correct integer number for the double difference hours for each station except for the station on Hans Island, ambiguity must be determined. This is not a trivial task, but which was only observed for 45 minutes. several ambiguity resolution techniques do exist, see for

163 In 2001 GPS data were available from 28 July to 24 August. mined, the final coordinates in the reference frame wanted for This fieldwork took place at the peak of the current solar the results can be then determined. More information on the activity cycle, but fortunately during a time period with low to GPS data processing is given by WEBER (2003). moderate ionospheric activity. The observation time spans in 2001 were between one and a half and three and a half days for each station. PRELIMINARY RESULTS OF THE PROCESSING

The 1995 data were initially processed right after completion In Figure 4 the RMS of repeatability is given for some of the of the survey campaign, but in order to have as similar condi- stations. RMS values are only determined for stations with tions as possible for the coordinate comparison, the data were more than one observation session, i.e. stations observed for reprocessed along with the 2001 data set. Both processing more than 24 hours. runs were carried out using the same processing parameters, and also, in order to have as similar conditions as possible, the The RMS values give an indication of the stability of the solu- differential GPS networks to be processed were designed to be tion. Generally the 2001 data set is better than the 1995 data. as identical as possible (see Figs. 2, 3). There are many small gaps in the data from 1995, probably caused by the older generation of equipment used, but also the The IGS stations at Thule Air Base have been used as the observation time intervals were shorter in 1995, and in some central reference stations, and have thus been constrained at cases really too short considering the length of the vectors. their given coordinates. For the 1995 data, station THU1 was Stations with only a few observation sessions are generally used, and the nearby and newer station THU2 was used for the more sensitive to noisy data, so both the shorter observation 2001 data. time spans and the increased amount of noise in the 1995 data translate into worse position accuracies and thereby the poorer For differential GPS positioning a higher accuracy is obtained repeatabilities shown in Figure 4. by using long observation time spans and short inter station vectors. In our case the distances from some of the stations in Generally the repeatability is about 1 cm in the horizontal, and the Nares Strait region to Thule Air Base are long, so in order 1.2 cm in the vertical for the 1995 data, and 3 mm in the hori- to obtain the best accuracy possible, the stations with the zontal, and 6 mm in the vertical for the 2001 data. In order to longer observation time spans were used for processing of the estimate the level of position accuracy, external error sources, long vectors. Vectors with short observation time spans could mainly centring of the GPS antennae, should also be consi- then in most cases be tied to stations located closer than the dered. In this case, special small tripods were used, and great stations at Thule Air Base. Hereby shorter GPS vectors, and care was taken in centring and in measurement of antenna thereby a better position accuracy, is obtained. height, and the contribution of antenna centring to the error budget is estimated to be 1-2 mm, at the most. For the data processing precise satellite positions, the final precise post-processed orbits from the IGS, were used in order to minimise the influence of errors in the satellite positions. FINAL GPS RESULTS The data were processed with a 15° elevation mask in order to eliminate the noisy data received at the lowest elevation For the final processing step both THU1 and THU2 were angles. The tropospheric error was mitigated using the Saasta- constrained to coordinates given in the ITRF97 epoch 15 moinen global tropospheric delay model (SAASTAMOINEN August 2001. When constraining the two Thule stations to 1973), and an ionosphere free linear combination of the obser- coordinates given at the same epoch in time any movement vations from the L1 and L2 frequencies was used to mitigate that might have taken place in the Thule area within the six the ionospheric effect. years is neglected, and Thule is thereby assumed to be a stable. This is of course not the case in reality, but the assumption The data were processed using the Bernese Software version makes it possible to analyse whether the entire Nares Strait 4.2 (BEUTLER et al. 2000). The processing software is consi- region is moving in the same way. dered to be one of the best and most advanced processing soft- wares, and the only known differential residual factors that are The general movements in the Thule area can be determined not handled by the software, and could have an influence on from the ITRF97 global GPS velocity solution, where Thule the positions, are ocean and atmospheric loading effects. Both has moved –10 cm in the North direction, 10 cm in the East are, however, expected to be negligible in the present case. direction and –2 cm in the Up direction over the six years.

The GPS data processing is carried out in several steps; first For any future observation campaigns the new station markers preliminary station coordinates are determined, then the ambi- (postfix 2 in the station name) will be used since they are more guities are resolved, and more precise station coordinates are stable than the old markers (postfix 1 in station name). The determined. For these initial processing steps it is important to new markers will thus be used for analysing the movements in avoid geometrical misalignments in the processing, and the the area and it is therefore necessary to generate 1995-coordi- coordinates used for the reference station must therefore be nates for these markers that were established in 2001. For this given in the exact same reference frame as the satellite posi- purpose the vector components obtained between the old and tions. The initial data processing is therefore carried out using the new markers from the processing of the 2001 data, are the ITRF93 reference frame for the 1995 data, and ITRF97 for used to generate 1995 coordinates for the new markers. The the 2001 data (BOUCHER & ALTAMIMI 1996). When the ambi- distances between the old and the new markers are between 15 guities are resolved and the vector components are deter- and 40 metres, and it is assumed that no local movements have

164 Fig. 4: RMS of repeatability for the coordinates from 1995 and 2001. taken place between the location of old and the new markers at and this is not sufficient for a vector of about 90 km. The 1995 the various sites during the six years. results for this station are therefore considered unreliable, and the station is not used for any further analysis in this paper. Final coordinates for 1995 and 2001, expressed as Northing, Easting and height above the reference ellipsoid, for all the For station WIL1 the differences are also large, about 7-9 cm stations used for the geodynamic analysis are shown in Table for each coordinate component. The RMS values (Fig. 4) are 1. only a few mm for the 2001 data, and the explanation for the large differences might thus be found in the first data set. In 1995 the observation time span was 24 hours for the 130 km COMPARISON OF 1995 AND 2001 CAMPAIGNS vector, and the observations were collected on June 3. When analysing the Northern Polar Cap Geomagnetic index a Coordinate differences between the two sets of coordinates slightly increased geomagnetic activity during the morning (Tab. 1) can now be determined. The coordinate differences, hours of the current day is found and this might have affected and thereby an indication of the movements that have taken the GPS signal propagation and thereby also have decreased place at the locations between May/June 1995 and August the positioning performance. The results for station WIL1 are 2001, are shown in Table 2. These coordinate shift values therefore also considered inaccurate, and will not be analysed should be seen in relation to the accuracy estimates (Fig. 4). any further. Generally the coordinate shifts in the horizontal are larger than the RMS of the coordinate values, indicating that the move- The apparent movements in the horizontal for the remaining ments are significant. For the height the results are generally stations are plotted in Figure 5, where the size and direction of more noisy. This is further discussed below. the 6-year coordinate shifts are indicated by the length and direction of the black vectors in the plot. There seems to be a The large differences for station HNS2 are caused by problems correspondence between the size and direction of the move- with resolution of the ambiguities in processing of the 1995 ments of stations ISA1, ALX1 and MAD2. However, station data. The observation time span in 1995 was only 45 minutes, JOE2 and BAI1 move in other directions, and the magnitude

Fig. 5: Horizontal movements 1995 to 2001, relative to Thule.

165 Tab. 1: Final GPS coordinates, UTM zone 22 and ellipsoidal heights, ITRF97 Fig. 6: Uplift (mm/yr) from the ICE-4G glacio-isostatic model.

numbers can be compared to contemporary glacial-isostatic uplift models, expressing the modelled land uplift due to histo- rical changes in ice load coupled with a rheological earth model. Figure 6 shows the uplift from the ICE-4G model (PELTIER 1996). It is seen that relative to Thule the central Nares Strait region is a region of high uplift, up to 5 mm per year, thus given a 3 cm effect over 6 years which – barely – should be detectable.

Figure 7 shows the GPS-observed relative height change over the 6-year period. The data are seen to be quite noisy, but at least the sign and order of magnitude of the movements are consistent with ICE-4G. It is therefore clear that on longer time spans (say, 10 year+), monitoring of uplift in this region by campaign-style GPS should have a lot of potential, especially considering the improved mounting and superior GPS observation quality of the 2001 campaign.

Tab. 2: Differences between station coordinates in 1995 and 2001. of JOE2 is much larger than the other stations, and could be affected by local effects. The overall pattern of the horizontal movements show that no major strike-slip movement exist along the Nares Strait, but otherwise results are consistent with a residual plate tectonic rotation relative to Thule.

Fig. 7: Observed and modelled land uplift for 1995-2001 for the roughly For the five “good” stations (Tab. 2) the movements in the South to North transect of “good” GPS stations along the Nares Strait, relative vertical direction are about 1-4 cm over the six years. These to Thule.

166 GPS AND SEA-LEVEL GEOID COMPARISON the major low in Kane Basin due to a deep regional mass deficit. The Nares Strait GeoCruise provided an opportunity to addi- tionally collect GPS positions linked to mean sea-level. These At two locations in Inglefield Land (Cape Alexander and Cape data are useful for evaluating the geoid models of the region Agassiz) temporary tide gauges were placed for a few days as for major systematic errors. The basic equation for the geoid part of the 2001 survey, and the tide gauge reference points height, N, derived from GPS and tide-gauge observations is were tied to GPS bolts onshore. Both a conventional portable tide gauge and a low-cost unit were tested. Figure 9 shows an NGPS = hGPS – Hsea-level - (2) example of the sea-level record at Cap Agassiz of the low-cost unit. With the relatively short duration of the sea-level obser- where h is the GPS ellipsoidal height, H the height above vations (2-3 days) only a rough mean sea-level can be esti- mean sea-level of a local tide gauge benchmark, and the mated, since long-period tides will produce aliasing. However, mean sea-surface topography. The latter is a small number for geoid control, accuracies of a few dm are sufficient to (sub-m) and likely rather constant in a small sea area like the control possible long-wavelength geoid errors. Nares Strait, and it is neglected in the sequel, together with the differences in the permanent tidal effects (GPS and geoid is referred to a tide-free system, whereas the mean sea level is by definition in the mean tidal system; this might yield an insi- gnificant error at the cm-level, much smaller than geoid errors).

The contours of the geoid of the Nares Strait region are shown in Figure 8. This geoid model is derived from a mix of satellite data, gravity data, and terrain models, as part of an Arctic- wide gravity field compilation, the “Arctic Gravity Project” (KENYON & FORSBERG 2001). The model should be reasonably good in the Nares Strait region, partly because of the 1995 gravity project. The features on the geoid are essentially corre- sponding to low-pass filtered gravity free-air anomalies, with Fig. 9: Tide-gauge observation at Cape Agassiz, as a function of relative time (hours).

Table 3 shows the elevation above sea-level results of the new tide gauge observations (denoted new) and older local mean sea-level observations, both given for nearby GPS points, locations of which are shown with crosses in Figure 8. Also in Table 3 geoid heights derived from GPS and sea-level observations, using Equation (2), are shown for the same points. Because the older points are tied in with triangulations up to some 10-20 km extent, the older values of sea-level height can be quite uncertain, which is indicated in the last column in Table 3 where the differences between the observed geoid

Tab. 3: Elevation above sea-level, derived geoid heights at the GPS points, and Fig. 8: Geoid heights (meter) in the Nares Strait region from the ArcGP geoid difference to ArcGP geoid model, derived from both new and older sea-level model. Crosses show location of GPS sea-level points. observations.

167 heights and the geoid heights from ArcGP are given. making it possible to carry out the geodetic project in connection with the Nares Strait GeoCruise. The 2001 survey was It is seen that Joe Island does not fit well, and indicates a gross done in close cooperation with Bob Morris (NRCan), and error in the old triangulation (the 1976 sea-level observation Lasse Nielsen (Asiaq - Greenland Survey). We thank NRCan was on the mainland of Greenland and not on Joe Island and Asiaq for the support of the project. itself). Also, at Alert the elevations seem not consistent, and the accuracy of connection from the benchmark 68B25 to sea- level is unknown. For the other points the consistency of the References data seems reasonably accurate, with the 20 cm RMS variabi- Beutler, G., Brockmann, E., Dach, R., Fridez, P., Gurtner, W., Hugentobler, U., lity being within the range of errors as possible in the ArcGP Johnson, J., Mervart, L., Rothacher, M, Schaer, S., Springer, T. & Weber, geoid, and also in the variability domain of the sea-surface R. (2000): Bernese GPS Software. Astronomical Institute, University of Berne. topography. The new points observed in 2001 fit well with the Bourcher, C. & Altamimi, Z. (1996): International Terrestrial Reference other points, and have complemented the older data with new, Frame.- GPS World, Advanstar Publ. 7/9: 71-74. more reliable, points and observations. Brunner F. K. & Gu, M. (1991): An improved model for the dual frequency ionospheric correction of GPS observations.- Manuscripta Geodaetica 16: 205-214. Forsberg, R., Ekholm, S., Madsen, B., Jensen, A. & Olsen, H. (1995): Gravity CONCLUSIONS and GPS measurements in Greenland 1995, Nares Strait (Op Bouguer 95) and East Greenland regions.- Survey & Processing Rep., Nation. Surv. Cadastre Denmark, 1-45. Due to the opportunity of the 2001 Nares Strait GeoCruise Han, S. & Rizos, C. (1997): Comparing GPS ambiguity resolution techni- seven GPS sites first observed in 1995 were revisited. The two ques.- GPS World, Advanstar Publ. 8/10: 54-56. GPS data sets from the stations have been processed and Jensen, A.B.O, Morris, R. & Nielsen, L. (2001): Cruise Report, Nares Strait Geo Cruise 2001, Geodetic Subproject.- In: R. JACKSON (ed), Passage analyzed in order investigate the geodynamical activity that 2001, pathway to the Artic, seismic survey and geoscientific experiment, has taken place in the area within the six years. Reliable results Geol. Surv. Canada (Atlantic), Bedford Inst. Oceanography, Dartmouth, were obtained for five of the stations, and all five stations 51-58. Kenyon, S. & Forsberg, R. (2001): Arctic Gravity Project – a status.- In: M.G. seem to be rotating relative to Thule. Vertical movements of Sideris (ed), Gravity, geoid and geodynamics 2000, Internat. Assoc. the stations are roughly coincident with contemporary glacial- Geodesy Sympos. 123: 391-395. isostatic rebound models, but for more reliable estimates Kleusberg, A. & Teunissen P.J.G. (eds) (1996): GPS for geodesy.- Lecture future campaigns should be carried out, utilising the new Notes in Earth Sciences, Springer-Verlag, Heidelberg, 1-407. Neilan, R.E., Zumberge, J.F., Beutler, G. & Kouba, J. (1997): The international improved benchmarks put in place 2001. GPS service: a global resource for GPS applications and research.- Proc. 10th Internat. Techn. Meeting Satellite Div. Inst. Navigation (ION GPS- With the additional observation of sea-level GPS heights, new 97), Kansas City, 883-891. Papitashvili, V.O., Gromova, L.I., Popov, V.A. & Rasmussen, O. (2001): Nort- reliable “GPS-levelling” data were added to the Greenland set hern polar cap magnetic activity index PCN: Effective area, universal of such data, allowing the calibration of geoid ties to mean-sea time, seasonal and solar cycle variations.- Sci. Rep. 01-01. Danish level, in principle defining the vertical datum surface. The data Meteorol. Inst., Copenhagen, 1-63. Peltier, W.R. & Jiang, X. (1996): Mantle viscosity from the simultaneous inver- along with older data indicate that the recent ArcGP geoid sion of multiple datasets pertaining to postglacial rebound.- Geophys. model seems quite accurate. Res. Lett. 33: 503-506. Saastamoinen, J. (1973): Contributions to the theory of atmospheric refraction.- Bull. Geodesique 105: 279-298, 106:383-397, 107:13-34. (printed in three parts). ACKNOWLEDGMENTS Seeber, G. (1993): Satellite geodesy.- Foundations, methods applications, Walter de Gruyter, Berlin, 1-589. We thank the BGR (Bundesanstalt für Geowissenschaften und Weber, M. (2003): GPS observations at Nares Strait.- Unpubl. proc rep., Rohstoffe) Germany (Dr. Franz Tessensohn) and the BIO, Nation. Surv. Cadastre Denmark, 1-10. Geological Survey of Canada - Atlantic (Dr. Ruth Jackson) for

168